Deep learning for Recommendations & Real-time results for image & video analysis

We're excited to invite you to another great meetup about DL for Recommendation systems & Online results for DL models in Image and video analysis hosted by Amdocs (aia unit)

Agenda:

17:30 – 18:00 – Gathering & Networking
18:00 – 19:00 - Deep learning for Recommender systems – Boris Rabinovich
19:00 - 19:20 - Pizza, beer and networking
19:20 – 20:30 - Online real-time results for image and video analysis - Elyasaf Korenwaitz

Both talks will be in Hebrew.

Logistics:
The meetup will take place at WOPA cafeteria ( HaSheizaf St 4, Ra'anana )
Parking – In front of WOPA building ( free after 17)

Deep learning for Recommender systems

Abstract:
We are going to explore a variety of deep neural networks to tackle the collaborative filtering (CF) problem.
We will concentrate on a major problem in recommendation systems - implicit feedback interactions.
Additionally, we will show how to improve an existing CF network by adding content data.
Finally, we will demonstrate how to plan a recommendation system experiment with evaluation protocols.

BIO:
Boris Rabinovich is a Data Scientist at Amdocs aia unit (AI). His experiences range from hands-on implementation of recommendation systems, a variety of NLP tasks,
predictive modeling for real-time-bidding systems, and constrained optimization to statistical process control and AB testing.
His work involves development of machine learning algorithms in a big data environment. He graduated from Ben-Gurion University with an M.Sc. in Information Systems Engineering.
His thesis project involved designing and developing a schema matching algorithm.

Online real-time results for image and video analysis

Abstract:
We are going to speak about how to use number of intensive models ("slow and heavy"), and combining many models in a real-time system,
while still receiving fast and precise results by efficiently using processing power.
Another topic is processing and analyzing videos in real time (less than a second to receive a result for a 1-hour video)
and deciding which class is in a video.

BIO:
Elyasaf is graduated from Open University with an BSc in Computer Science.In the Hi-Tech industry since 2005,Team Leader since 2006, R&D Manager since 2008, and CTO since 2010 at NetSpark.
Six registered patents in the field of classification and processing of signals, texts, images and videos.

Patents:
1.On-line video filtering (pending and not yet published)
2.(US20140095515A1) Real-time single-sweep detection of keywords and content analysis
3.(US9529896B2) Hierarchical online-content filtering device and method
4.(US9805280B2) Image analysis systems and methods
5.(US20120041326A1) Method and system for measuring heart rate variability
6.(US20130079652A1) Assessment of cardiac health based on heart rate variability